Broker due diligence for a self-trader

Tier-one regulation is the entry condition. FCA, FINMA, ASIC, BaFin, the SEC, and the major CySEC firms operating under EU passport are the jurisdictions we will consider. We do not trade with offshore brokers, regardless of the marketing. This is a floor — passing it does not make a broker safe, it just makes them eligible.

Beyond the licence itself we read the audited financial statements where available, look at the regulator's enforcement history against the firm, and read the firm's resolution arrangements. A broker with a clean licence and segregated client funds is meaningfully safer than one with a clean licence and no segregation, even though both pass the superficial check.

Execution quality cannot be assessed from marketing materials. We run a standardised order set on a small live account for a defined period — typically a month — before scaling any capital onto the venue. The order set covers market orders, limit orders, stop orders, partial fills, and behaviour during scheduled high-impact news events.

Demo accounts are useless for this. Brokers do not run demo and live infrastructure identically, and the slippage profile is almost always better on demo. The only data we trust comes from real fills on real capital, however small.

What we are looking for is not just average slippage. We are looking for the tail. A venue with low average slippage and a fat tail of bad fills during news is operationally worse than a venue with slightly higher average slippage and a thin tail. The tail is where the real cost lives.

Brokers are excellent at accepting deposits. The harder question is what happens when you ask for a meaningful withdrawal. We test this quietly, at full size, before any venue earns trust. A broker that processes a small token withdrawal in a day but takes three weeks on a real one is not a venue we keep.

Withdrawal track record is one of the few non-negotiable signals. We have seen otherwise excellent brokers fail this test and have closed accounts accordingly. We have also seen modest brokers pass it cleanly and earned long-running relationships with them as a result.

Regardless of how excellent a broker looks, we never concentrate more than a defined fraction of working capital at a single venue. Brokers fail. Sometimes they fail because of fraud, sometimes because of regulatory action, sometimes because of operational incidents that are nobody's fault. The cause does not matter to the account that was holding the capital.

Multiple smaller relationships beat one large one, every time. The operational overhead is real — more accounts, more reconciliations, more relationships to manage — and it is a price we pay deliberately. The supervision layer enforces venue concentration limits in real time alongside the other risk limits.

The publicly available broker comparison sites optimise for retail concerns — minimum deposit, demo account quality, app design, copy-trading features. None of that is what matters for a self-trading book of meaningful size. We read those sites for entertainment, not for due diligence.

The questions that actually matter are not in the rating tables. How does the broker behave when a major counterparty defaults? What is the firm's exposure to a single liquidity provider? Has the firm ever widened spreads punitively during a stress event? These questions are answered in conversations with the firm's institutional desk, in court records, and in the experience of other professional traders — not in star ratings.